Report top-5000 primes here

[michaf]

Well.... they MIGHT be at 79,80,81,82 and 83k respectively...
I MIGHT be off by some amount too...

Don't give up :)

[gd_barnes]

Originally Posted by Siemelink

Bingo!

7127*48^78407-1 is prime with 131825 digits. It comes in at place 2505 in the top 5000 list. I still need 5 more primes to prove the conjecture for prime riesel conjecture for base 48, so chances are that proving it is out of reach for now.

Cheers, Willem.

Congrats on a large prime Willem!

[gd_barnes]

Originally Posted by Siemelink

Bingo!

7127*48^78407-1 is prime with 131825 digits. It comes in at place 2505 in the top 5000 list. I still need 5 more primes to prove the conjecture for prime riesel conjecture for base 48, so chances are that proving it is out of reach for now.

Cheers, Willem.

Willem,

Even though you stated it, it took me a while to figure out that you were actually working on the "prime" riesel conjecture for this base. At first, I thought you had the wrong base.

Unfortunately I won't be able to show the prime on the web pages as they currently exist since the "regular" conjecture is only k=3226.

At some point, I'll create separate pages for the prime conjectures and make them a sub-project. Of course I'll show the prime at that point.


Gary

[mdettweiler]

Originally Posted by gd_barnes

Willem,

Even though you stated it, it took me a while to figure out that you were actually working on the "prime" riesel conjecture for this base. At first, I thought you had the wrong base.

Unfortunately I won't be able to show the prime on the web pages as they currently exist since the "regular" conjecture is only k=3226.

At some point, I'll create separate pages for the prime conjectures and make them a sub-project. Of course I'll show the prime at that point.


Gary

Maybe in the meantime, we could keep track of the "prime" conjectures in a table inside a dedicated thread?

[Siemelink]

Originally Posted by mdettweiler

Maybe in the meantime, we could keep track of the "prime" conjectures in a table inside a dedicated thread?

The easiest would be to drop my excel somewhere on Gary's website, coupled with a link from the main page. Add a note to it that this is the Dec 1 version and ready for now.
I'll start on double checking at some point, when that is done I'll post the latest version.

Willem.

[mdettweiler]

Found a nice big juicy one for Sierp. base 23!:

8*23^119215+1 is prime!

This is also my first top-5000 prime for a non-power-of-2 base.

[gd_barnes]

Originally Posted by mdettweiler

Found a nice big juicy one for Sierp. base 23!:

8*23^119215+1 is prime!

This is also my first top-5000 prime for a non-power-of-2 base.

Outstanding!! I don't have a single top-5000 prime on a non-power-of-2 base yet. Nice work!

This now becomes one of an amazing 7 bases <= 32 on the Sierp side that only have ONE k remaining! Those last k's are about as stubborn as they come. All except your k=68 have been searched to at least n=195K.


Gary

[gd_barnes]

Originally Posted by mdettweiler

Maybe in the meantime, we could keep track of the "prime" conjectures in a table inside a dedicated thread?

Be my guest. I have all that I can handle now, especially since KEP gave up the Riesel base 3 conjecture. I still haven't reviewed the recent efforts on it to include them in my pages yet.


Gary

[MrOzzy]

Originally Posted by gd_barnes

Outstanding!! I don't have a single top-5000 prime on a non-power-of-2 base yet. Nice work!

This now becomes one of an amazing 7 bases <= 32 on the Sierp side that only have ONE k remaining! Those last k's are about as stubborn as they come. All except your k=68 have been searched to at least n=195K.


Gary

I don't know how the people who reserved them feel about it, but what about combining all these bases and k into a single big sieve with afterwards a big drive for it to take them all to n= 1 million or so? (using llrnet for example)
The chance one of the bases gets proven is rather high if you do it this way ...

[mdettweiler]

Originally Posted by MrOzzy

I don't know how the people who reserved them feel about it, but what about combining all these bases and k into a single big sieve with afterwards a big drive for it to take them all to n= 1 million or so? (using llrnet for example)
The chance one of the bases gets proven is rather high if you do it this way ...

Well, LLRnet doesn't work with multiple different bases/types at once, so we'd still have to use separate servers for each base. Probably the best way would be to hold team drives on each base in turn, and then when one is proven, move on to the next.

[gd_barnes]

Originally Posted by MrOzzy

I don't know how the people who reserved them feel about it, but what about combining all these bases and k into a single big sieve with afterwards a big drive for it to take them all to n= 1 million or so? (using llrnet for example)
The chance one of the bases gets proven is rather high if you do it this way ...

It's interesting that you mention that. I recently have been having visions of running CRUS a little like NPLB in some respects.

This could actually be done but we wouldn't want to do it with LLRnet because Phrot is likely 30-50% faster than LLR and as Max said, LLRnet can't handle more than one base at a time.

What we would have to do is create a big ABC file with multiples bases in it that can be fed into either PFGW or Phrot. That way, many bases could be tested at once.

That said, the specific bases are at such widely varying test depths, ranging from n=100K to 400K that we couldn't really do it to prove these particular conjectures per se until we evened them out a little more, which would take quite a while. But we COULD do such a thing on k's for mutliple bases in whatever other manner that we choose as a team effort. Perhaps both base 19's at once since they are both at n=25K. Or several of the bases > 32 that are currently at n=10K all at once.

There are myriads of possibilities.

All of that said, due to many things going on at NPLB at the moment, I wouldn't want to coordinate such an effort for several months yet.

Before embarking on such an endeavor though, what we really need to do is have a major team discussion about some goals for CRUS. We don't really have any other than the little "mini-goal" that I made to get all of our powers-of-2 bases up to n=600K base 2 by year end. Right now, we're just kind of searching willy-nilly here-and-there because us admins (namely this one) haven't really had the time to put some direction to the project.


Gary